CN112328651B - Traffic target identification method based on millimeter wave radar data statistical characteristics - Google Patents
Traffic target identification method based on millimeter wave radar data statistical characteristics Download PDFInfo
- Publication number
- CN112328651B CN112328651B CN202011115125.0A CN202011115125A CN112328651B CN 112328651 B CN112328651 B CN 112328651B CN 202011115125 A CN202011115125 A CN 202011115125A CN 112328651 B CN112328651 B CN 112328651B
- Authority
- CN
- China
- Prior art keywords
- data
- target
- millimeter wave
- traffic
- wave radar
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2465—Query processing support for facilitating data mining operations in structured databases
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/21—Design, administration or maintenance of databases
- G06F16/215—Improving data quality; Data cleansing, e.g. de-duplication, removing invalid entries or correcting typographical errors
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2462—Approximate or statistical queries
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/28—Databases characterised by their database models, e.g. relational or object models
- G06F16/284—Relational databases
- G06F16/285—Clustering or classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
- G06F18/24133—Distances to prototypes
- G06F18/24137—Distances to cluster centroïds
- G06F18/2414—Smoothing the distance, e.g. radial basis function networks [RBFN]
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Abstract
The invention discloses a traffic target identification method based on millimeter wave radar data statistical characteristics, which comprises the following steps: collecting traffic target data by using a millimeter wave radar; noise data cleaning processing is carried out, and data sets of different traffic targets are constructed; dividing each traffic target data set into a training set, a verification set and a test set; performing correlation analysis on each attribute of the radar data to find out the attribute with high correlation with the target category; performing statistical analysis on the obtained radar data with the attributes to construct empirical characteristics; training a classifier by using the empirical characteristics, the obtained attributes and a training set to obtain a target recognition model; and identifying the category of the target to be detected by using the high model. According to the invention, a traffic target recognition model with higher practical value is established through less detection data analysis and modeling. And secondly, the traditional radar target recognition task is combined with machine learning algorithms such as SVM (support vector machine), LSTM (least squares metric) deep learning models and the like, so that the recognition precision is improved, and the method has high adaptability.
Description
Technical Field
The invention belongs to the technical fields of artificial intelligence recognition algorithms, millimeter wave radar sensors and the like, and particularly relates to a traffic target recognition method based on millimeter wave radar data statistical characteristics.
Background
Driven by the popularity of the internet of vehicles (IoV) and by marginal computing, many autonomous driving technologies have evolved rapidly. In order to prevent and reduce the occurrence of traffic accidents, it is necessary to warn the driver of pedestrians on the road in real time by using pedestrian recognition technology. Pedestrian vision inspection techniques have been extensively studied. However, severe weather severely affects the detection performance of the vision sensor. The millimeter wave radar sensor has strong impact resistance to environmental influences caused by severe light and weather conditions. Therefore, the millimeter wave radar-based pedestrian recognition technology can improve the performance of advanced driver assistance systems. In the past, target identification is mainly performed from radar data of traffic targets at a single sampling moment, and the motion situation of the targets is considered to be less, so that a relatively high identification accuracy rate is not obtained finally, and the method has no practical use value. Moreover, conventional radar data target identification tasks require the establishment of large target detection databases to better identify targets.
Disclosure of Invention
The invention aims to provide a traffic target identification method based on millimeter wave radar data statistical characteristics, aiming at the problems in the prior art.
The technical solution for realizing the purpose of the invention is as follows: a traffic target identification method based on millimeter wave radar data statistical characteristics comprises the following steps:
step 1, collecting traffic target data by using a millimeter wave radar;
step 2, carrying out noise data cleaning processing on the acquired data to construct data sets of different traffic targets; dividing each traffic target data set into a training set, a verification set and a test set;
step 3, performing correlation analysis on each attribute of the radar data, and finding out the attribute of which the correlation with the target category is greater than a preset threshold value;
step 4, performing statistical analysis on the radar data with the attributes obtained in the step 3 to construct an empirical characteristic;
and 6, collecting radar data of the target to be detected, inputting the data into a target recognition model, and outputting the category of the target.
Further, the collecting traffic target data by using the millimeter wave radar in the step 1 specifically includes: and the millimeter wave radar is utilized to collect data of vehicles, pedestrians and non-motor vehicles in a static state, a moving state and a turning state.
Further, the attributes related to the target category and greater than the preset threshold in step 3 include the distance, the speed and the radar reflected energy value RCS value of the target.
Further, the empirical feature in step 4 is DRCS:
in the formula, vx is the velocity component in the parallel direction, vy is the velocity component in the vertical direction, and RCS is the radar reflection energy value.
Compared with the prior art, the invention has the following remarkable advantages: 1) a large amount of historical data is not needed for model training, and the requirement on the real-time performance of target recognition is met; 2) the proposed model has strong practicability and short prediction time; 3) the method has the advantages that the parameter difference among different targets is fully understood, and the recognition accuracy of the algorithm is further improved by designing the target recognition model.
The present invention is described in further detail below with reference to the attached drawing figures.
Drawings
FIG. 1 is a flow chart of a traffic target identification method based on millimeter wave radar data statistical characteristics according to the present invention.
Fig. 2 is a schematic view of an application scenario of the present invention.
Fig. 3 is a distribution diagram of raw radar data for different targets.
Fig. 4 shows RCS values and statistical distributions of stationary pedestrians at different distances, where fig. 4 (a) shows the RCS values and fig. 4 (b) shows the statistical distributions.
Fig. 5 shows RCS values and statistical distributions of pedestrians walking at constant speed at different distances, where fig. 5 (a) shows the RCS values and fig. 5 (b) shows the statistical distributions.
Fig. 6 shows RCS values and statistical distributions of pedestrians turning at different distances, where (a) of fig. 6 shows the RCS values and (b) of fig. 6 shows the statistical distributions.
Fig. 7 shows RCS values and statistical distributions of different-distance stationary vehicles, where fig. 7 (a) shows the RCS values and fig. 7 (b) shows the statistical distributions.
Fig. 8 shows the RCS values and the statistical distribution of the vehicles traveling at constant speed for different distances, where fig. 8 (a) shows the RCS values and fig. 8 (b) shows the statistical distribution.
Fig. 9 shows RCS values and statistical distributions of vehicles turning at different distances, where (a) of fig. 9 shows the RCS values and (b) of fig. 9 shows the statistical distributions.
FIG. 10 is a graph of classification accuracy of different machine learning algorithms on radar data with newly proposed features added.
Fig. 11 is a graph showing the classification effect of the LSTM network on 3s time series data, where fig. 11 (a) is a graph showing the training accuracy and fig. 11 (b) is a graph showing the training loss.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.
In one embodiment, in combination with fig. 1, a traffic target identification method based on millimeter wave radar data statistics is provided, the method includes the following steps:
step 1, collecting traffic target data by using a millimeter wave radar, wherein the traffic target data comprises a target distance, a target speed, a target direction, an RCS value and the like;
step 2, carrying out noise data cleaning processing on the acquired data to construct data sets of different traffic targets; dividing each traffic target data set into a training set, a verification set and a test set;
preferably, 80 percent of the data in the data set is used as a training set, 10 percent of the data is used as a verification set, and 10 percent of the data is used as a test set.
Step 3, performing correlation analysis on each attribute of the radar data, and finding out the attribute of which the correlation with the target category is greater than a preset threshold value;
step 4, performing statistical analysis on the radar data with the attributes obtained in the step 3 to construct an empirical characteristic;
and 6, collecting radar data of the target to be detected, inputting the data into a target recognition model, and outputting the category of the target.
Further, in one embodiment, the collecting traffic target data by using the millimeter wave radar in step 1 specifically includes: and the millimeter wave radar is utilized to collect data of vehicles, pedestrians and non-motor vehicles in a static state, a moving state and a turning state. Illustratively, the data collected is as shown in table 1 below:
table 1 example of data collected
Further, in one embodiment, the attributes of step 3, which are related to the target category and greater than the preset threshold, include the distance, the speed, and the radar reflected energy RCS value of the target.
Further, in one embodiment, the performing the statistical analysis in step 4 specifically analyzes the variance, the mean and the distribution of the radar data.
Further, in one embodiment, the empirical characteristic in step 4 is DRCS:
in the formula, vx is the velocity component in the parallel direction, vy is the velocity component in the vertical direction, and RCS is the radar reflection energy value.
Here, the new feature proposing process is as follows:
through the statistical distribution and the correlation analysis of the measured data, a new characteristic DRCS is constructed in the form of a mathematical formula. According to the distribution and correlation of the data, the following factors are considered to represent the difference between different objects:
1) the distance between the radar sensor and the object.
2) The velocity of the object in x and y spatial coordinates.
3) The direction of the object motion.
4) The original RCS value of the object.
The acquired radar data for each target includes the range, velocity and RCS values of the target. These parameters, although having a certain recognition effect on the recognition of vehicles and pedestrians, can be obtained in the following analysis. However, different target data may overlap to some extent, which makes the identification accuracy of the original radar data alone not very high.
Illustratively, the statistical analysis of the collected radar data is performed according to different motion types, and the results are shown in fig. 3 to 9. As can be seen from the figure, there is a large amount of overlap between the data.
The RCS of a vehicle is generally higher than a pedestrian, but with more overlap. The highest RCS of the pedestrian is well below the vehicle. When the RCS is greater than 10, the vehicle can be considered.
At normal speeds, the vehicle's RCS is generally greater than the pedestrian's RCS, but the difference is not very large. There is a large amount of overlap. The speed between the vehicle and the pedestrian is very different and therefore the speed is given a high weight when deriving new features. Moreover, the faster the speed, the heavier the weight.
When the vehicle turns, the vehicle body part fluctuates greatly during turning, and the severe dragging phenomenon appears on two sides of RCS distribution.
Based on the above analysis, the DRCS features proposed by the present invention are obtained.
Further, in one embodiment, the classifier in step 5 comprises a Support Vector Machine (SVM) or a long-short-term neural network (LSTM).
An application scenario of the invention is shown in fig. 2, for example, recognition effects of multiple machine learning methods such as a support vector machine are shown in fig. 10, and recognition effects of a long-term neural network are shown in fig. 11. According to the invention, the traffic target recognition model with higher practical value is established through less detection data analysis and modeling. And secondly, the traditional radar target recognition task is combined with machine learning algorithms such as SVM (support vector machine), LSTM (least squares metric) deep learning models and the like, so that the recognition precision is improved, and the method has high adaptability.
The foregoing illustrates and describes the principles, general features, and advantages of the present invention. It will be understood by those skilled in the art that the present invention is not limited to the embodiments described above, which are given by way of illustration of the principles of the present invention, but that various changes and modifications may be made without departing from the spirit and scope of the invention, and such changes and modifications are within the scope of the invention as claimed. The scope of the invention is defined by the appended claims and equivalents thereof.
Claims (5)
1. A traffic target identification method based on millimeter wave radar data statistical characteristics is characterized by comprising the following steps:
step 1, collecting traffic target data by using a millimeter wave radar;
step 2, carrying out noise data cleaning processing on the acquired data to construct data sets of different traffic targets; dividing each traffic target data set into a training set, a verification set and a test set;
step 3, performing correlation analysis on each attribute of the radar data, and finding out the attribute of which the correlation with the target category is greater than a preset threshold value;
step 4, performing statistical analysis on the radar data with the attributes obtained in the step 3 to construct an empirical characteristic; the empirical characteristics are DRCS:
in the formula, vx is a velocity component in the parallel direction, vy is a velocity component in the vertical direction, and RCS is a radar reflection energy value;
step 5, training a classifier by using the empirical characteristics, the attributes obtained in the step 3 and a training set to obtain a target recognition model;
and 6, collecting radar data of the target to be detected, inputting the data into a target recognition model, and outputting the category of the target.
2. The traffic target recognition method based on millimeter wave radar data statistical characteristics according to claim 1, wherein the collecting of the traffic target data by the millimeter wave radar in the step 1 is specifically: and the millimeter wave radar is utilized to collect data of vehicles, pedestrians and non-motor vehicles in a static state, a moving state and a turning state.
3. The method for identifying traffic targets based on millimeter wave radar data statistical characteristics according to claim 1 or 2, wherein the attributes of the correlation with the target category larger than the preset threshold in step 3 comprise the distance, the speed and the radar reflection energy RCS value of the target.
4. The method for identifying traffic targets based on millimeter wave radar data statistical characteristics as claimed in claim 3, wherein the statistical analysis in step 4 is specifically to analyze the variance, mean and distribution of radar data.
5. The method of claim 1 wherein the classifier in step 5 comprises a Support Vector Machine (SVM) or a long-short-term neural network (LSTM).
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011115125.0A CN112328651B (en) | 2020-10-19 | 2020-10-19 | Traffic target identification method based on millimeter wave radar data statistical characteristics |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011115125.0A CN112328651B (en) | 2020-10-19 | 2020-10-19 | Traffic target identification method based on millimeter wave radar data statistical characteristics |
Publications (2)
Publication Number | Publication Date |
---|---|
CN112328651A CN112328651A (en) | 2021-02-05 |
CN112328651B true CN112328651B (en) | 2022-09-06 |
Family
ID=74314011
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202011115125.0A Active CN112328651B (en) | 2020-10-19 | 2020-10-19 | Traffic target identification method based on millimeter wave radar data statistical characteristics |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112328651B (en) |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109932720A (en) * | 2019-03-19 | 2019-06-25 | 北京润科通用技术有限公司 | Motion estimate method and device based on radar |
CN111583671A (en) * | 2020-06-05 | 2020-08-25 | 南京信息职业技术学院 | Millimeter wave radar intersection traffic flow monitoring method and system |
-
2020
- 2020-10-19 CN CN202011115125.0A patent/CN112328651B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109932720A (en) * | 2019-03-19 | 2019-06-25 | 北京润科通用技术有限公司 | Motion estimate method and device based on radar |
CN111583671A (en) * | 2020-06-05 | 2020-08-25 | 南京信息职业技术学院 | Millimeter wave radar intersection traffic flow monitoring method and system |
Also Published As
Publication number | Publication date |
---|---|
CN112328651A (en) | 2021-02-05 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Wang et al. | A comparative study of state-of-the-art deep learning algorithms for vehicle detection | |
CN110942000B (en) | Unmanned vehicle target detection method based on deep learning | |
CN105892471B (en) | Automatic driving method and apparatus | |
US8818702B2 (en) | System and method for tracking objects | |
Shi et al. | Fast and robust vanishing point detection for unstructured road following | |
CN110738690A (en) | unmanned aerial vehicle video middle vehicle speed correction method based on multi-target tracking framework | |
CN102270301A (en) | Method for detecting unstructured road boundary by combining support vector machine (SVM) and laser radar | |
CN108230254A (en) | A kind of full lane line automatic testing method of the high-speed transit of adaptive scene switching | |
CN112487905B (en) | Method and system for predicting danger level of pedestrian around vehicle | |
Li et al. | An adaptive 3D grid-based clustering algorithm for automotive high resolution radar sensor | |
CN115630569A (en) | Millimeter wave radar sensor modeling method applied to automatic driving | |
CN106327528A (en) | Moving object tracking method and operation method of unmanned aerial vehicle | |
CN114359876B (en) | Vehicle target identification method and storage medium | |
Ye et al. | Overhead ground wire detection by fusion global and local features and supervised learning method for a cable inspection robot | |
CN104915642A (en) | Method and apparatus for measurement of distance to vehicle ahead | |
CN112810619A (en) | Radar-based method for identifying front target vehicle of assistant driving system | |
Xiao et al. | Traffic sign detection based on histograms of oriented gradients and boolean convolutional neural networks | |
Qing et al. | A novel particle filter implementation for a multiple-vehicle detection and tracking system using tail light segmentation | |
CN114299607A (en) | Human-vehicle collision risk degree analysis method based on automatic driving of vehicle | |
Fries et al. | Combining model-and template-based vehicle tracking for autonomous convoy driving | |
CN104463909A (en) | Visual target tracking method based on credibility combination map model | |
Chen et al. | Vehicle detection based on multifeature extraction and recognition adopting RBF neural network on ADAS system | |
CN104200226A (en) | Particle filtering target tracking method based on machine learning | |
Kim et al. | Semantic point cloud-based adaptive multiple object detection and tracking for autonomous vehicles | |
CN112328651B (en) | Traffic target identification method based on millimeter wave radar data statistical characteristics |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |